In 2008, colorectal cancer (CRC) was the fifth most prevalent cancer in the USA accounting for 149,800 new cases and the second most deaths, ~49,000. Progression of CRC normally occurs over decades and involves a combination of selection and the accumulation of selective gene alterations. A large body of work examining gene mutations in human colorectal tumor specimens strongly suggests that progression from a normal epithelial cell to a metastatic adenocarcinoma is a multistep process requiring multiple mutations in genes such as APC, K-RAS and TP53. However, the constellation of gene mutations varies considerably and a single combination is not seen in all colorectal cancers. This variability in cancer genotypes confounds interpretations because simple progression models although instructive fail to fit all tumors. Certain key questions remain unanswered: What is the minimal combination of gene mutations sufficient for propelling a normal epithelial cell into a malignant adenocarcinoma? Does the overall order of mutation accumulation matter or can some mutations occur at random? Ideally, one would like to take a prospective approach and engineer different stage tumors in an intact organism. This is one goal of the present proposal, using genetically engineered mice. Unlike many models in which the mouse is heterozygous for a given mutation or the gene is conditionally mutated in all cells within the target tissue, we will model intestinal tumorigenesis by stochastic, somatic alteration of one or more cancer genes in widely separated cells in vivo. We developed a highly versatile stochastic Cre/lox system to target specific gene mutations to individual cells in the mouse. Cre activation occurs by spontaneous frameshift reversion, can be modulated by DNA mismatch repair status and occurs in individual cells surrounded by unaltered (normal) tissue, mimicking spontaneous cancer causing mutations. Cell lineages that experience Cre activation also now express 2-galactosidase, therefore illuminating the consequences of tumor suppressor gene inactivation and oncogene activation. As one primary goal, we will study the short-term and longer-term consequences of individual tumor suppressor gene or oncogene alterations with our stochastic system. As a second goal, with our system we will study mice with multiple gene targets to define the minimal gene (pathway) disruption(s) that are sufficient for invasive intestinal adenocarcinomas. Tumors arising in mice with different target combinations will be analyzed in some detail to provide additional information on the intestinal tumorigenesis pathway(s). These studies are intended to provide a better system for modeling and thereby understanding human colorectal cancer pathway(s).